[FLINK-38814][Connector/Kinesis] Kinesis Source in EFO - Fix idle consumer on subscription failure #224
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- call void error(String var1, Throwable var2) instead void error(String var1, Object... var2)
…shardSubscriber for thread-safety
he FanOutKinesisShardSubscription could enter a zombie state where it appeared active but was not consuming events. This occurred due to incorrect handling of the `ResourceInUseException` during the subscription phase. Problem: When the Kinesis SDK threw a `ResourceInUseException`, the code incorrectly released the `waitForSubscriptionLatch`. This signaled "success" to the waiting thread, even though no `onSubscribe` callback had occurred. Root Cause: 1. The waiting thread woke up and attempted to call `requestRecords()` on a null `subscription` object, throwing a `NullPointerException`. 2. The `CompletableFuture.runAsync` lambda only caught `InterruptedException`. The `NullPointerException` (a RuntimeException) killed the thread silently. 3. The `subscriptionActive` flag remained true, causing the main poll loop to wait indefinitely for events that would never arrive. Changes: - Updated the `.exceptionally()` block to treat `ResourceInUseException` as a failure (calling `terminateSubscription`) rather than a success. - Added a catch-all `catch(Throwable t)` block to the async activation thread to ensure unexpected errors are logged and handled rather than swallowed. - Added a safety check to ensure `subscription` is not null before requesting records.
nadberezny
changed the title
nadberezny
changed the title
|
java.util.concurrent.TimeoutException: Timeout when subscribing to shard shardId-000000000033 with starting position StartingPosition{shardIteratorType=AFTER_SEQUENCE_NUMBER, startingMarker=49670023483232461788730612813735157811693693431491592722} can this exception also be related to this? I have a similar use case described in the issue, in which the consumer gets in a zombie like state |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Purpose of the Change
This PR fixes a critical stability issue in the FanOutKinesisShardSubscription where the connector could enter a "zombie" state—appearing active but not consuming events—following a failed subscription attempt (ResourceInUseException).
It also resolves some potential duplicates due to non thread safe shard starting position.
More details can be found in the JIRA ticket description:
https://issues.apache.org/jira/browse/FLINK-38814